Au. ‘01

Version 1.1

Tests whether means of different groups are equal. An analysis of variance is for discrete independent variables: there must be just a few values of the variable; to every value of the independent variable must correspond a group of participants who have that value. An analysis of variance is used if there are more than two groups. If you compare just two groups, you use a t-test.

E.g.: If you want to know whether students taught with three different teaching methods (=independent variable) have the same mean grade or not, you would use an analysis of variance.

Applied research

Research that has immediate use, that has a practical goal.

Basic research

Research that is mostly targeted at finding out abut reality, and at testing theories. Basic research does not have immediate practical goals

Case

One research participant, or one firm analyzed in your research, or one country. In general: the unit of your research

Case study

Study in which one case is thoroughly analyzed

Chi-square test

A test of whether frequencies are as expected. There can be lots of different expectations that can be tested.

Example 1: Are there more or less women in a class that men? With a chi-square test, you can test the expectation that the frequency of maleness in a class = the frequency of femaleness = 50 %.

Example 2: Are there more animal lovers in England that in France? With a chi-square test, you can test the expectation that the percentage of animal lovers in England = the percentage of animal lovers in France.

Condition

Group of participants who gets treated in the same way in an experiment. E.g.: in an experiment about how good aspirin works against headache, the participants who get the aspirin form one condition (the experimental condition), and those that do not get aspirin but instead a placebo are the other condition (the control condition).

Construct

Theoretical variable, something you cannot directly measure. E.g.: power is a construct, a variable that is used in theories, but that is difficult to measure. Intelligence is a construct (you measure it with IQ-test, but the IQ does not exactly measure intelligence.

Correlation

Relation between two variables expressed in a number. A correlation can be between –1 and 1.

A correlation of 1 means that if the one variable is high, than the other variable is also high, and vice versa. In other words, there is an extremely strong relation between the two variables.

A correlation of 0 means that if one variable is high, than the other variable is as often high as it is low. In other words, there is no relation between the two variables.

A correlation of -1 means that if the one variable is high, than the other variable is always low, and vice versa. In other words, there is an extremely strong, but negative relation between the two (the two variables are opposites of each other).

Experiment

Study in which the level of the independent variable is manipulated. The experimenter creates conditions in that way, that are then compared to each other on the dependent variable. E.g.: in an experiment about how good aspirin works against headache, the participants who get the aspirin form one condition (the experimental condition), and those that do not get aspirin but instead a placebo are the other condition (the control condition). The two conditions are then compared on how much head ache they have one hour after taking a pill (the dependent variable).

General linear model.

A model that assumes that all scores on a variable are the sum of the effects of other variables.

E.g.: your score on an exam = effect of the time you spent studying + effect of your intelligence + effect of your motivation + effect of your mood on the day you were tested + effects of all other variables not yet mentioned.

Hypothesis

A conjecture, an idea from the researcher about how things might be in reality. You can test a hypothesis in a study

Interview

Asking a participant questions face-to-face. Can be structured, so that all questions are written out in advance, or unstructured, so that the researcher talks with the participant without a preset plan.

Journal

Scientific magazine, in which researchers can publish their results. It usually selects its articles through peer review

Level of measurement

Precision of measurement. Can be on nominal, ordinal, interval or ratio level.

Nominal measurement: there are just categories, without any order. E.g.: political party

Ordinal: there are categories, but they are ordered. Calculating a mean does not make sense E.g.: scales of earthquake strength

Interval: the differences between the units are meaningful, and you can calculate an average. You can change the whole scale if you wanted to, because there is no real zero. E.g.: temperature, where zero celsius = freezing point of water is just a convention and other conventions are also possible

Ratio: the differences between units is meaningful, and there is a real zero. E.g.: most quantities, such as weight, height, the number of things, etc. You an change the measurement scale of weight, but 0 kilos will always remain 0 kilos.

Manipulation

Setting the level of an independent variable, controlling it

Matching

A strategy to make sure that two groups of participants are similar, by making sure that their score on some measure is equal. E.g.: in a study about the effect of university on high school students, you can select your high school students in such way that the intelligence of those that go to university is equal to that of those that don’t go to univeristy (I.e., you match the two groups on intelligence)

Mean

Average. You calculate it by summing all the values and then divide by the number of cases. E.g.: the mean of 4, 5 and 6 is: 4 + 5 + 6 = 15 / 3 = 5 (divided by three because I gave three numbers).

Measurement

Measuring a variable means putting a value on every case according to a standard procedure. E.g.: measuring length by putting the participant against the wall, making a stripe above his/her head an that looking with the measurement stick how high the stripe is.

Observation

Looking at the participant while he or she does things. Usually the behaviour of the participant is scored on some things. E.g.: you can look at children playing, and score how well they interact during play.

Participant

A human who participates in your research (a subject of your research)

Peer review

A way to select articles (usually referred to as ‘papers’) that will get published in a journal, or proposals that will get funded. It involves asking some other researchers in the field of the researcher who wrote the paper or the prosoal, to make an anonymous judgment on the paper or proposal. In essence, it means that your colleagues in science determine if your work is good or not.

Population

All cases that could be part of your research. E.g.: if you are interested in K.S.U. students, your population is 12,000 students. If you would want that your research is valid for all Kosovars, then your population would consist of all 1,000,000 or so Kosovars. If you would like to do research that is valid for all of humanity (nearly all psychological research wants this), then your population consists of 5,000,000,000 people.

Qualitative research

Research without numbers and precise measurement, and usually without a detailed plan in advance (there is room for improvisation, and the reaction of the participants help shape the research. Usually participant observation, or open interviews, or case studies.

Quantitative research

Research in which numbers are used, usually totally laid out in advance. Usually Surveys or experiments.

Regression

A technique to predict one variable with the help of another. You can use regression with two continuous variables. Only if the variables are related (i.e., if there is a correlation between the two), can you use regression to predict one of the two with help of the other.

E.g.: length and weight are related, so you can use length to predict weight. If someone is 1.75, you can predict that he or she will weigh 75 kilos. That does not mean that the person will always weigh 75 kilos, but you can predict better if you know the length than if you don’t.

Reliability of measurement

How well can you repeat your measurement? If you repeat it, does it then give you the same score?

Sample

Group of participants in your research. Your sample is always a subset of the research population. E.g.: if you want to do research on K.S.U. students, you have a population of 12,000 (the number of students). If you do research on just 20 of them, that is your sample.

Scale

Set of measurements that together form one single measurement

Standard deviation

The square root of the variance. Often the standard deviation is reported instead of the variance because it is of the same size as the measurements themselves (while the variance is usually much bigger because you square numbers to calculate it), and because a certain percentage of people (+- 68%) is between one standard deviation below the mean and one standard deviation above the mean

E.g.: with IQ the mean of the whole population is 100. The variance is 225. Therefore the standard deviation is Ö 225 = 15. Now you know that 68% of the people have an IQ between 100-15 = 85 and 100+15 = 115.

Subject

See participant

Survey

Study in which questions are asked, either through a questionnaire or through a structured interview, to a large sample. The main goal of a survey is to generalize the answers to the questions of the sample to the population. E.g.: an opinion poll is a survey, but also student evaluations after a course.

T-test

Tests whether the means of two groups are equal or not (if there are more than two groups, you must use an analysis of variance). The dependent variable has to be continuous. You can also use a t-test to test whether a mean of some continuous variable is bigger or smaller than some value (for example, if the mean rainfall in Pristina in a month is bigger or smaller than 90 mm.). E.g.: a t-test can be used to test whether the concentration of students (as measured by their test score) is bigger if they sit in confortable chairs (group one) than if they sit on wooden chairs (group two).

Validity of measurement; construct validity

Answer to the question: does your measurement measure what it is supposed to measure?

Validity of research: internal and external

Internal validity means: is the conclusion you draw from your research really true? Is your explanation of your results the only possible one? External validity means: how well does your conclusion generalise from your sample to the whole population

Value

What one case has on one variable. E.g.: the value of the case Martijn Meeter on the variable length is 1 meter, 97 centimeters

Variable

Something that can vary between cases. E.g.: some of the variables on which humans differ is hair calour, length, weight, gender, political preference, salary, how often we brush our teeth, etc.

Variance

A measure of how far from the mean all the cases are. You calculate it by taking the difference between every value and the mean. Then you square that difference, sum those squares and divide the sum by the number of cases. E.g., say that you have as scores: 5, 6, and 10. Then the mean is: 5+6+10 = 21 / 3 = 7. Then you calculate the variance by first calculating the squared difference between each value and the mean (the mean=7):

5 - 7 = -2; -2 squared = 4

6 - 7 = -1; -1 squared = 1

10 - 7=3; 3 squared = 9

So the variance is: 4 + 1 + 9 = 14 / 3 = 4.666666

Ethics

Field work