Glossary of Terms

Glossary (A to Z) of psychological terms used in
A Guide to Coursework in Psychology by Eamon Fulcher.

To search for a term
Press Ctrl+F (the control key Crtl and the F key at the same time) then enter the search term.

You can report any errors or omissions to me if you wish here.

Alternative hypothesis
The experimental hypothesis that predicts a statistically significant result.
It is the opposite to the null-hypothesis, which predicts a non-significant result.

Anecdotal evidence
Everyday observations of behaviour that appear to be consistent with a psychological explanation.
Anecdotal evidence (of a theory) is not really evidence at all and cannot be used by itself to support a theory.

APA referencing style
The recommended way of referencing previous work in a psychological report or essay.
APA stands for the American Psychological Association.

Appendices (of a report)
The end sections of a report that contains information about an investigation that is not fully illustrated in the text.
Information that appears in an appendix is usually things like the complete set of instructions given to participants, examples of materials, such as photographs or lengthy passages of text, questionnaires and a smaple of completed questionnaires, and so on.

Bar chart
A graph which represents values (usually group means) as vertical bars.
See Exercise 2 >>

Between-subjects design
An experimental research design in which a sample of participants is divided into two groups on the basis of random selection.

Carryover effects
An effect that can occur with a repeated measures design in an experiment, where the participant's experience in the condition carried out first can affect their behaviour in a second condition.
For example, if the two conditions require the learning of a new skill, then performance on the second condition is likely to benefit from practice in the first condition. One way to minimise this for each participant is to have a preliminary practice session before participating in either condition. One way to minimise this within a sample of participants is to alternate the order in which participants are tested, i.e., half do condition 1 then condition 2, while the other half do condition 2 then condition 1.

Categorical scale
A scale of measurement that consists of categories, such as sex, birth place, type of sport one enjoys playing, and so on.

Cause and effect
A relationship between two things, where one thing makes a change happen to another thing. Experimental psychology is about trying to identify cause and effect relationships.
An experiment could be said to be an attempt to recreate a cause and effect relationship by manipulating the conditions under which it could be observed (by manipulating the independent variable and seeing its effects, if any, on the dependent variable).

Ceiling effect
An effect that occurs when a task is too easy for all or most of the participants.
When a task is too easy, most scores will be high or at the maximum possible score. A good task is one which yields a good spread of scores.

Chance
Something that occurs randomly.
Clearly, if everything has a cause and effect then there can be no such thing as chance! However, in statistics it refers to the likelihood that any numerical difference observed between two conditions is NOT the result of the IV on the DV but has been caused by something else that hasn't been measured.

Chi-square test
An inferential test that compares the spread of scores that would be expected to occur by chance with the actual spread that has been observed. If the observed scores vary sufficiently from scores expected by chance then the result is statistically significant. Whether the scores vary sufficiently from chance is calculated by the test.
This test is used on data that is categorical or nominal (when you are looking at the proportions of participants who fall into different categories.)

Computed P
The actual computed or cacluated probability that the obtained data (e.g., the difference between two mean, or the correlation between two variables) could have occurred by chance.
Inferential statistics are used to calculate p and p is usually compared against the stated alpha value.
See Exercise 1 >>

Confounding variable
Something that can obscure or interfere with the cause and effect relationship being observed in an experiment.
In an experiment it is desirable for there to be only one difference between two groups on the independent variable. However, if one group is affected by something other than the independent variable (such as a sudden outside noise or by a different instruction) but not the other group then this is a confounding variable. This confound could either improve or just change the behaviour of one group and hence could be the cause of differences in behaviour rather than the independent variable.

Consent form
A form given to participants prior to their participation that requests their permission to take part in the investigation.
The idea of a consent form is to raise the awareness that participation is voluntary and to form a contract between the experimenter and participant about the conduct of the experiment.

Content analysis
A research method whereby some material (typically text) is analysed statistically (such as by counting the occurrences of particular words. See the example in the book).

Continuous scales
A scale of measurement where a value can be anything that is within a range of values (and the recorded value is usually rounded, e.g., to 2 decimal places). For example, length is a continuous scale. Contrast this with a discrete scale, such as 2, 4, 6, 8, 10, where a value can only be one of these numbers.

Contrived observation
An observation carried out in a pre-arranged location, such as a psychology lab.
Often the investigator may be unseen by the participants, by viewing the activities through a one-way mirror.

Control group
A group of participants who are given an identical task as the experimental group except for one feature. This single feature is present in the task given to the experimental group but not in the task given to the control group.

Controls
Experimental controls are the steps taken to identify and deal with (in advance of the study) any potential confounding variable.

Correlation, negative
An 'inverse' relationship between two variables, such that for a group of participants as their scores increase on one variable their scores decreases on another variable.
A negative correlation is expressed as a number between 0 and -1. For example, a correlation of -0.75 is a strong negative correlation.

Correlation, positive
A direct relationship between two variables, such that for a group of participants as their scores increase on one variable their scores also increase on another variable.
A positive correlation is expressed as a number between 0 and +1. For example, a correlation of 0.62 is a strong positive correlation.

Correlation, zero
The lack of any relationship between two variables, such that for a group of participants it is not possible to say anything about their scores on one variable even if we know their scores on the other variable.
A zero correlation is expressed as a number close to 0, such as -0.1 or +0.1. A correlation that is found to be non-significant (regardless of its actual value) is often considered the same thing as a zero correlation, i.e., there is no association between the two variables.

Correlational design
A research design where participants are measured on at least two variables (on an interval or ratio scale) and a test of correlation is carried out on the data of the two variables to determine whether there exists any form of relationship between them.
A correlational design can be similar in many ways to an experiment in that it can involve participants carrying out a laboratory task; however, there is no independent variable and no dependent variable.

Cross-sectional design
A research design usually used to investigate the effects of time. Participants of different age groups are tested on some measure or task to determine the effects of ageing on the measure.

Counterbalancing
A way of controlling confounding variables, such as order effects.

Critical value
A probability value obtained from statistical tables that indicates the value of the obtained test statistic needed in order for the result to be deemed statistically significant.

Dependent variable
In an experiment, this variable is measured from all participants and is the main task that is given.
Differences between the experimental group and the control group on the dependent variable can reveal an effect that is caused by the single difference between the two groups.

Dichotomous scales
A scale of measurement where a value can be either one thing or another (e.g., 0 or 1).

Directional hypothesis
A hypothesis that predicts a difference or relationship that specifies the direction of the difference (e.g., that the mean response of the experimental group will be larger than the mean of the control gorup rather than the other way around) or relationship (e.g., that there will be a positive correlation between two variables rather than a negative correlation).

Discourse analysis
A research method that does not involve the use of experimental manipulation or statistics but focuses on conversations and text in an attempt to establish how people construct their reality or views of the world.
See an outline of Discourse Analysis in 'deleted scenes' from the book>>

Error variance
The scores on a variable can vary between participants. When this variance is not ude to the independent variable it is error variance, and is a source of unreliable variation that cannot be explained.
Error variance can be due to measurement errors, the method of data collection, and so on.

Ethical guidelines
A set of guidelines that investigators should follow in order to protect participants from physical or psychological harm and also to protect their identity.
Ethical guidelines are produced by the British Psychological Association.

Ethnographic research
Ethnography is a research method concerning the link between culture and behavior. It involves describing the details of social life or cultural phenomena in a small number of cases.

Exam board
A meeting of a group of people employed by an examination service to ensure that exam papers are set, marked and graded appropriately.

Experimental group
A group of participants who are given an identical task as the control group except for one feature. This single feature is present in the task given to the experimental group but not in the task given to the control group.

Experimental hypothesis
A precise prediction about the way the results will turn out in the experiment.
An experimental hypothesis may predict that the scroes from one group will be significantly higher than the scores of another group, or it may predict a positive or negative correlation between two variables.

Face validity
A form of validity of a questionnaire or construct, that is not the direct result of a statisical analysis, but one that is a qualitative judgement made by examining the contents of the questions with reference to what is known about the construct from research.
The face validity of a measure of anti-social personality disorder, for example, could be assessed by comparing the set of questions against what is known about the disorder from research, and especially case studies.

Field experiment
An experiment not carried out in a laboratory but in a more natural setting.
For example, a study of bystander behaviour towards a acted out event in the street would be a field experiment.

Flat distribution
A fequency distribution, where the frequencies of each level are similar and hence the graph appears flat and without peaks and troughs.

Floor effect
An effect that occurs when a task is too difficult for all or most of the participants.
When a task is too difficult, most scores will be very low. A good task is one which yields a good spread of scores.

Frequency
The number of times something occurs.
For example, if 8 participants scored between 90% and 100% on a memory task, then the frequency of scores in the interval 90-100% would be 8.

Frequency distribution
A graph of the levels of a variable against the frequency of each level.
For example, a frequency plot of scores on a memory test could consist of a graph of the number of participants that scored each possible score of the test.

Generalisation
In an experiment, a generalisation of the results occurs when one applies the findings found in a sample of participants to the population.
For example, the findings from a study involving a small number of anxious students might be generalised to all anxious individuals. If the study found that the more anxious participants recalled more in a memory test than did less anxious participants, it may be tempting to conclude that anxiety causes improvements in memory. In this way, a generalisation can be seen to be the abstraction of a rule about behaviour from a single study.

Grouped frequency distribution
A way of simplifying a frequency distribution by grouping the scores on a variable into groups or classes.
In a simple frequency distribution, one would plot the frequency of every possible score. However, when there is a large number of possible scores, many may have a frequency of 0. In this case the scores can be grouped. For example, if scores ranged from 0 to 100%, one could group them into 10 units, such as 0-9%, 10-19%, 20-29% and so on.

Histogram
A bar chart of a grouped frequency distribution.

Hypothesis
A precise statement that makes a prediction about the results of a study. The purpose of the study is to test this prediction.
A hypothesis is an element of a theory. A theory should consist of at least one testable prediciton. An experiment usually consists of an experimental hypothesis and a null-hypothesis.

Hypothesis testing
This refers to the fact that an investigation, and especially an experiment, is an exercise in hypothesis testing; thus, the purpose of an investigation is to test a hypothesis.

Independent groups design
An experimental design that consists of two separate groups of participants performing an identical task, except for one difference (i,e., the difference that is defined by the independent variable).

Independent variable
The element of an experimental design that defines the single difference between the way two groups of participants are treated or the difference between two experimental conditions.
For example, if we wish to study the effects of the colour of printed words (red or black) with a white background on reading spead, we could devise two conditions that are identical except for one element, namely the colour of the printed words. So, participants in the experimental condition, read a passage of text printed in red, and the control group read the same passage of text and under the same conditions but with the words printed in their usual form, black. The only difference between the two groups is the colour of the printed words. Any differences in reading speed (in this case the dependent variable) between the two groups will be caused by the colour of the printed words (assuming that both groups are identical in every other way). In practice, there will always be some differences between the two groups, even if it just the fact that the two groups are comprised of different people.

Inference
In quantitative psychological research, the drawing of a conclusion about whether the data supports or refutes the null-hypothesis.

Inferential statistics
The type of statistics used to help draw inferences about the data gathered in an investigation.

Inferential test
A particular type of test of the family of statistics known as inferential statistics, that is used to help draw a conlusion about whether the data gathered in an investigation supports or refutes the hypothesis.

Information sheet
An outline of the investigation, given to the participants before they agree to participate in the study.
The aim of the information sheet is to help participants decide whether or not to participate in the study. It should be detailed enough, in terms of the procedure, the materials, and the type of behaviour being requested, to help them make an informed decision.

Interbehaviour latency
A measurement of the time between two occurrences of the same behaviour.
This measure is usually taken, amongst others, in obervational studies, where such a factor may be important for the hypothesis being tested.

Interval scale
A scale of measurement where the intervals on the scale are equal.
An obvious example is length in metres - the difference between 1 and 2 metres is exactly the same as the difference between 6 and 7 metres. An example of something that is NOT an interval scale is position in a race (the difference between 1st and 2nd could be one second and the difference between 4th and 5th could be one minute).

Journal article
A journal article is an essay or research report published in a volume of other, similarly themed articles.
A journal consists, usually, of volumes that are numbered annually, and issues that a re numbered monthly, bimonthtly or quarterly. A journal issue usually contains about 8 articles. An article is either a research report of an investigation or an essay-type review of research in a particular area. Articles a peer reviewed before they can get published, whcih means that only articles that satisfy a set of strict criteria will be accepted for publication.

Laboratory experiment
An experiment carried out in a laboratory, rather than in a naturally occurring setting, because it is easier to control the situation (and especially extraneous or confounding variables).

Literature review
An essay-type body of text that summarises an area of research, usually with the primary aim of justifying or providing a rational for an investigation.
The introduction to a report usually begins with a literature review that summarises research and theory that are direclty relevant for the current investigation. A good literature review is one that provides an accurate background of theory and research to the current study, and identifies weaknesses or gaps in theories that the present study will address.

Longitudinal design
A research design where the same measure is applied to the same sample of participants at (at least) two points in time, such as before and after an important event or simply after a set interval (e.g., 5 years).

Mann-Whitney U test
An inferential test that compares the scores from two groups and computes the probability that any difference is due to chance.
This non-parametric test is used when the data are ordinal or interval, for a test for differences, and with an independent measures design. There do not have to be equal numbers of participants in the experimental and control groups.

Matched groups design
An experimental design that consists of two groups of participants that are matched on some attribute, such as intelligence.
In an unrelated groups design, where two groups of participants are used, the participants may differ between the two groups on some aspect that is highly pertinent to the topic of study. If one group has more of this ability then scores from that group may be higher than scores from the other group, regardless of the independent variable (i.e., the thing that is supposed to be the only difference between the two groups). One way of controlling for this is to match the two groups on that ability, such that the mean and spread of scores on that ability are very similar between each group. It should be noted that a matched group design does not attempt to make both groups as identical as possible in many attirbutes, but only in one or possibily two abilites.

Mean
The mean is the numerical average and it is usuallu calculated for each group or each condition.

Measures of central tendency
A measure of central tendency is one which is supposed to describe some central feature of a set of scores, such as the mean, median, or mode.
Note that each of these can be misleading and it is always wise to present a measure of the spread of scores, such as the standard deviation.

Measures of spread
A measure of how the scores in a set of scores vary, such as the range, variance, or standard deviation.
The most commonly used measure of spread is the standard deviation and can be thought of as roughly equal to the average distance of all scores from the mean.

Median
The middle score of a set of scores.

Median split
A way of dividing a set of participants into two groups on some measure by finding the median score and allocating those participants with scores higher than the median to the 'high' group and those with scores lower than the median to the 'low' group.
The contemporary view is that the method of the median split is logically flawed (see the book for more details).

Missing data
Data that is incomplete for reasons such as recording errors, participants giving an inappropriate response, or participants giving a response that is difficult to decipher.
It does not have to be catastrophic when a participant has one or two missing scores from a task as there are ways with dealing with their overall scores (see the book for details).

Mode
The most commonly occurring score in a set of scores.

Naturalistic observation
A type of investigation in which observations are carried out in settings where the investigator does not intervene or interfere with the situation.

Nominal scale
A scale of measurement that consists of categories, such as sex, birth place, type of sport one enjoys playing, and so on.

Nondirectional hypothesis
A hypothesis that predicts a difference or a relationship but does not specify the direction of the difference or relationship.
This is the same thing as a two-tailed hypothesis, where the direction of the difference or association is unspecified.

Non-parametric test
An inferential test that is not dependent on a number of specific assumptions about the population from which the sample data is obtained.
For example, the data does not need to be drawn from a normally distributed population and the data does not need to be interval.

Normal distribution
A distribution of scores (1) that is symmetrical around the mean, (2) where the mean, median, and mode are the same, and (3) where the plotted frequency distribution appears bell-shaped.
When these assumptions are satisfied (even approximately) it is possible to make use of more advanced inferential statistics that make use of the known statistical properties of the normal distribution.

Norms
The norms of an inventory or published questionnaire refer to the previously researched scores obtained from specific groups of the population.
For example, a questionnaire can be applied to different types of workers, age groups, clinical versus non-clinical samples, and so on, and their data (means and standard deviations) will be included in the manual that accompanies a published questionnaire or inventory. It is then possible to use these norms to better understand scores obtained from participants in a new investigation.

Null hypothesis
A precise statement that makes the opposite prediction about the results of a study as the experimental hypothesis.
The purpose of the study is to test this prediction. More specifically, the inferential statistics used test the 'truth' of the null hypothesis.

Observational design
A type of research design where the investigator observes more naturally occurrin gbehaviour than is typical of a laboratory study. Observations made are usually recorded on tally sheets where specific actions are counted.

One-tailed hypothesis A hypothesis that predicts a difference or a relationship and also the direction of the difference or relationship.
Whereas a nondirectional or two-taile dhypothesis does not predict a direction, the one-tailed hypothesis should clearly state which group is expected to get the higher score (or quicker reaction time and so on) and in the case of a correlation, whether the expected correlation should be positive or negative.

Opportunity sample
A sample of participants for an investigation based on the sampling principle of 'opportunity', that is participants that were available through the easiest method, usually by the investigator being in one place and asking for volunteers.
Alternative methods include sampling specific populations, such as people with dyslexia, or throught he more complicated method of stratified sampling, where the sample should reflect the population on a number of measures, such as ethnic grouping, male-femal ratio, age range, and so on.

Order effect
An effect that can be present in a repeated measures design and occurs when all participnats undergo the same task before undergoing a second task.
Clearly, experience with the first task will affect experience with the second task, and may either improve or hamper performance on the second task. One way of combating or controlling for order effects is to mix trials of task 1 with trials of task 2 or by dividing participants into two groups and getting both groups to do the tasks in a different order.

Ordinal scale
A scale of measurement that consists of items in some logical order, such as position in a race, order of preference, a list of participants ordered by their academic achievements.

P value
The probability that the null hypothesis is true, calculated using inferential statistics.
The statistical test returns a P value that can be compared with an acceptbale or expected level (alpha).

Parametric assumptions
Certain assumptions about the population of scores from which a sample is taken. These are (1) the plotted frequency distribution is symmetrical around the mean, (2) the mean, median, and mode are the same, and (3) the plotted frequency distribution appears bell-shaped.

Parametric test
An inferential test based on a number of assumptions about the population from which the sample data is obtained.
One assumption is that the data is drawn from a normal distribution, another is that the data scale is at least interval.

Participant observation
A type of investigation using the observational method but where the investigator becomes an accpeted member of the group being studied.

Participants
Volunteers who have agreed to take part in an investigation.

Plagiarism
The deliberate or sometimes accidental (but only accidental in that the person doing the plagiarising is ignorant of the issue of plagiarism) copying of another person's work (either a published article or the work of anohter student) into one's own work and where the author of the original work is not acknowledged.
The term 'work' here can refer to any amount of material, from just one specific sentence to a whole report. See the book for some examples of plagairism and also how to avoid it.

Population(s)
The total group of items or persons that could be measured.
Population here does not necessarily mean the same thing as the population of a country or even the entire population of the world, but could mean a specific group of peaople or obejcts (such as the opulation of university lecturers in the UK, or the population of professional footballers living in London, and so on).

Practice effect
A type of order effect that can occur in a repeated measures design in an experiment, where experience with one task improves or informs performance on a subsequent task.

Probability
The likelihood of something.
For example, if the probability that it is going to rain this evening is 50%, then the likelihood that it will rain this evening is 50-50. The probability of an event can be calculated in specific ways; an inferential test is one that calculates the probability that the null hypothesis is 'true'. Usually probability makes an 'all else being equal' assumption, i.e. that things that might influence the outcome of an event that we cannot measure or account for (or even know exists) has, in fact, no influence at all!

Psychology journals
A psychology journal consists, usually, of volumes that are numbered annually, and issues that are numbered monthly, bimonthtly or quarterly. A journal issue usually contains about 8 articles. An article is either a research report of an investigation or an essay-type review of research in a particular area. Articles are peer reviewed before they can get published, which means that only articles that satisfy a set of strict criteria will be accepted for publication.

Quasi-experimental design
A research design that differs from an experimental design because participants are not randomly allocated to groups.
An example might be testing two classes from the same age group or workers from two different offices. There is the assumption that both groups are equivalent but this is unlikely to be the case. This method is most common in social psychology research.

Qualitative method
A method of research that does not involve the gathering of numbers or the statistical analysis of data. The data that is studied is usually text or conversations.

Quantitative method
A method of research that involves the gathering of numbers and the statistical analysis of data.

Questionnaire
A set of questions that have the same response scale (such as Yes-No, or Agree-Agree a little-Not sure-Disagree a little-Disagree).
Questionnaires need to go through a relaibility and validity check before they are widely adopted by other researchers. Usually a questionnaire is an inventory or a measure of some psychological characteristic (e.g., an attribute or attitude).

Random allocation
Allocating participants into two or more groups on the basis of random selection.
This is an important feature of an independent groups or unrelated experimental design. There is the assumption that the groups will tend to even out in terms of their overall psychological attributes (if there is a sufficient numbers of participants).

Random sample
A way of recruiting a fixed number of participants from a pool of potential participants by choosing them on some random basis (e.g., drawing their names out of a hat).
It is important to sample randomly, unless a certain criterion needs to be met. If one chooses in a systematic way (e.g., every pupil sitting at the front of the class) then the sample may be biased.

Ratio scale
A scale of measurement that consists of measurements that have a clear (absolute) zero or minimum point and where the intervals are of equal measure.
For example, length is a ratio scale because it has a minimum value of zero and the difference between say, 10 metres and 12 metres is the same as the difference between 1045 metres and 1047 metres.

Reaction time
The time taken to respond to a cue, usually measured in milliseconds (ms; where 1000 ms is 1 second).
When measuring reaction time it is nearly always the case that participants are asked to respond as quickly as they can but as accurately as they can (i.e., to avoid errors and hence missing data).

Reactivity
The state a participant may be in either when they know they are being observed or when they think they know the underlying hypothesis in an experiment (also called demand awareness).
A consequence of this state is that they may react or behave differently to how they would behave without such an awareness.

Recall
To remember something, usually over a short space of time.
Free recall is where participants are given an amount of time to remember specific items. Cued recall is where participants have to recall an item given a cue. Recall can be measured as the number of items recalled divided by the total number of items, and then multiplied by 100 to yield a percentage recall score.

References (of a report)
The section of a report that lists in alphabetical order the articles cited in the main body of the report.
There is a recommended way of referencing articles in a psychological report or essay and this way is published by the American Psychological Association.

Related design
An experimental research design in which a sample of participants all undergo more than one condition or treatment.
A related design is the same thing as a repeated measures design.
relative frequency distribution The percentage of participants who scored within a certain range of scores in a frequency distribution.

Reliability
The extent to which a measurement is accurate in the sense that the same measurement made on two occasions should produce the same value.

Interitem reliability
The correlation of responses to items on a questionnaire.

Slit-half reliability
The extent to which the responses from two halves of the same questionnaire correlate with each other.

Test-retest reliability
The correlation obtained between responses from a questionnaire when participants are tested on two occasions.

Repeated measures design
An experimental research design in which a sample of participants all undergo more than one condition or treatment.

Replication
The repeating of an experiment by following the procedure as closely as possible.
Replications are important for showing that an effect is a reliable one.

Representative sample
A sample of participants that is representative of the population on a number of attributes.

Response set
The production of the same response despite differences in the stimuli presented. This can happen in a questionnaire when, for example, the participant ticks "Strongly Agree" to every question. This effect can be reduced when a proportion of the questions are reversed (see the book for more details).

Rotation
A method of counterbalancing materials to reduce practice effects (see the book for more details).

Scattergraph
A visual representation of a correlation that consists of the scores on two variables from all observations.

Scientific approach
An approach to understanding something that involves carrying out investigations that are carried out to a strict set of criteria.

Sign test
A non-parametric test that computes the probability that two sets of scores differ significantly.

Significance tables
Numeric tables on which one compares a test statistic with values computed to be significant at vaious levels of alpha. Statistical tables are used when one is unable to compute actual significance values using a computer.

Skewed distribution
A distribution of scores that is asymmetrical, and the mean, median, and mode are towards the upper end or lower end of a scale.

Spearman's rho
A statistical test of correlation when the data satisfies parametric asummptions.
standard deviation A measure of the spread of scores that can be thought of as the average deviation of each score from the mean value.

Standardised instructions
A set of instructions given to all participants in a study.
In order to avoid confusion about what is required or misunderstanding in participants, an investigator needs to devise a set of instructions that are concise, very clear and not open to ambiguity. All participants shoud receive exactly the same instructions (unless, of course two different instructions are used as the independent variable).

Statistical significance
The probability that the null hypothesis is true, calculated using inferential statistics. For example, if an inferential test of differences yields the result P = 0.02, then this means that the calculated probability that the difference between two sets of data is due to chance (and not due to the independent variable) is 0.02 (which is the same thing as 1 in 50 or 2%).

Structured recording
An approach to making observations, where what is to be recorded is determined beforehand. Compare this with unstructured recording.

Summary data
Numbers, such as the mean and standard deviation, that are used to summarise or describe sets of data. Summary data is calculated by using descriptive statistics.

Surveys
A type of questionnaire that samples opion or beliefs on an issue. Unlike a questionnaire, a survey can use more than one response scale.

Tally sheet
A recoding sheet used by an observer in an investigation, where specific observed responses of behaviours are counted or tallied.

Test value
A numericall value produced by calculations using an inferentiall statistical test. Given the test value, the number of groups one has and the number of participants, it is possible to calculate the probability of the null-hypothesis.

Tests of significance
Inferential statistical tests that can be used to determiine whether the null hypothesis is "true".

Trend
This refers to the status of the result of a statistical test that produces a p-value marginally greater than alpha. In this case, the result can be said to be a trend towrds significance. For example, if alpha is 0.05 and the obtained p-value of a test was 0.08, then this value can be referred to as a trend. Although the result is not statistically significant it can be mentioned in a report since if the hypothesis is an important one and there are recognised methodological problems in the type of study being carried out, it suggests that a minor change might yield a significant result in a future replication.

t-test
A parametric statistical test for determining whether the difference between two sets of scores is significant.

Two-tailed hypothesis
A hypothesis that predicts a difference or a relationship but does not specify the direction of the difference or relationship. For example, the hypothesis may predict that the mean score of one group will be significantly different to scores of a second group, but it will not say which mean will be higher.

Type I error
A type I error is a FALSE HIT - the error of inferring that the experimental hypothesis is true when in "reality" it is false.

Type II error
A type II error is a MISS - the erros of inferring that the experimental hypothesis is false when in "reality" it is true.

Unrelated design
An experimental design that consists of two separate groups of participants performing an identical task, except for one difference (i,e., the difference that is defined by the independent variable).

Unstructured recording
An approach to making observations, where what is to be recorded is not completely determined beforehand but made "on the hoof". This is espcially useful when the relevant behaviours needed to be measured cannot be fully known before the study commences.

Validity
An assessment of whether something actually measures what it is supposed to measure.
variables A measurement that can take on different values. Height is a variable because it varies from one person to another.

Variance
A measure of differences between individual scores in a set of scores.

Wilcoxon matched-pairs signed ranks test
An inferential test based on non-parametric assumtpions and when the data are ordinal or interval, the hypothesise predicts differences, and when the design is repeated measures.

Within-subjects design
An experimental research design in which a sample of participants undergo more than one condition or treatment.