Search Holah

Investigations


 

 

 

 

Here are some exam style questions.

 

 

 

Here is a tick off what you need to know sheet for experiments.

 

 

 

 

 

Check out this matching quiz on variables.

 

This uber-test checks out your knowledge of IVs and DVs
A confounding variable is a variable which has an unintentional effect on the dependent variable. When carrying out experiments we attempt to control extraneous variables, however there is always the possibility that one of these variables is not controlled and if this effects the dependent variable in a systematic way we call this a confounding variable.

 

An extraneous variable is a variable which could effect the dependent variable but which is controlled so that it does not become a confounding variable.

 

A laboratory is any environment where variables can be well controlled. Such environments are usually artificial but do not have to resemble a science lab at school or college.

 

Control groups are often used in experiments. This is a group which does not receive the manipulation of the independent variable and can be used for comparison with the experimental group or groups.

The term ecological validity refers to how well a study can be related to or reflects everyday, real life. Studies with high ecological validity can be generalised beyond the setting they were carried out in, whereas studies low in ecological validity cannot.

 

Ethics are a set of guidelines which psychologists carrying out research should follow.

 

According to the ethical guidelines participants should be protected. That is, the experimenters should avoid psychological harm such as embarrassment, a loss of self esteem and changing a person. Participants should not experience any greater risks than they would encounter in their everyday lives.

 

Test yourself here.

 

 

 

 

 

 

 

 

 

 

 

Although field experiments do not have such tight control over variables they do have the advantage of being far less artificial than laboratory experiments.

 

 

 

 

 

A true experiment involves the deliberate manipulation of an independent variable by an experimenter. Therefore quasi or natural experiments are not true experiments as the independent variable is not directly manipulated. The researcher takes advantage of a naturally occurring independent variable.

 

Therefore natural experiments are also known as quasi experiment ? they are like experiments but are not true experiments.

 

 

Here is a match the type of experiment quiz.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

When a quasi or natural experiment is being carried out (such as when we are comparing males and females) because the participants cannot be randomly assigned to one of the conditions an independent measures design has to be used. Obviously the participants cannot be randomly assigned to one of the conditions as the condition is the quality of the participants (such as male or female).

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Here is an old school print out and fill in the gaps thing.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Here is a match the design quiz and here is another one.

 

 

 

 

 

 

 

Here is a cloze exercise on hypotheses for experiments.

 

 

Operationalising variables

 

When conducting experiments it is important to operationalise the variables. That is, stating a clear way that the independent variable is going to be manipulated and the dependent variable is to be measured.

 

For example if an experiment was to be carried out to see if time of day affected memory it would be important to operationalise the variables of time of day and memory. We might operationalise time of day as 10am and 10pm and operationalise memory as performance on a memory task.

 

In fact, it may be worth being even clearer with the operationalising of the dependent variable and state how the performance on a memory task is to be measured. For example number of words recalled out of a list of fifty words.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Another multi-choice quiz about the experimental method.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Nominal data are data in separate categories such as number of runners that finished a marathon.

 

Ordinal data are data that are ordered such as the order of runners that finished the marathon ? first, second, third and so on.

 

Interval data are data that are measured using a public unit of measurement. For example this could be the times in which the runners finished the marathon.

 

Home > Investigations > Experiment

Experiment

Experiment for Psychological Investigations

An experiment is a research method used by psychologists which involves the manipulation of variables in order to discover cause and effect. It differs from non-experimental methods in that it involves the deliberate manipulation of one variable, while trying to keep all other variables constant.

The two main types of 'true' experiments are laboratory experiments and field experiments.

When psychologists carry out experiments they use one of three basic experimental designs to investigate the effects of an independent variable on a dependent variable. These are the independent measures design, the repeated measures design and the matched pairs design.

Before researchers carry out experiments they operationalise the variables and create hypotheses. A hypothesis is a testable, predictive statement.

Experiments produce quantitative data which can be analysed statistically. For this part of the course you need to be aware of descriptive statistics including measure of central tendency and bar charts.

 

There are three main types of experiment - laboratory experiments, field experiments and quasi (natural) experiments.

 

Laboratory experiments

A laboratory experiment is an experiment conducted under highly controlled conditions.

The variable which is being manipulated by the researcher is called the independent variable and the dependent variable is the change in behaviour measured by the researcher.

 

All other variables which might affect the results and therefore give us a false set of results are called confounding variables (also referred to as random variables).

By changing one variable (the independent variable) while measuring another (the dependent variable) while we control all others, as far as possible, then the experimental method allows us to draw conclusions with far more certainty than any non-experimental method. If the independent variable is the only thing that is changed then it must be responsible for any change in the dependent variable.

Laboratory experiments allow for precise control of variables. The purpose of control is to enable the experimenter to isolate the one key variable which has been selected (the independent variable), in order to observe its effect on some other variable (the dependent variable); control is intended to allow us to conclude that it is the independent variable, and nothing else, which is influencing the dependent variable.

However, it must also be noted that it is not always be possible to completely control all variables. There may be other variables at work which the experimenter is unaware of.

It is argued that laboratory experiments allow us to make statements about cause and effect, because unlike non-experimental methods they involve the deliberate manipulation of one variable, while trying to keep all other variables constant.

 

Sometimes the independent variable is thought of as the cause and the dependent variable as the effect.

 

Furthermore, experiments can usually be easily replicated. The experimental method consists of standardised procedures and measures which allow it to be easily repeated.

 

However laboratory experiments are not always typical of real life situations. These types of experiments are often conducted in strange and contrived environments and the participants mat be asked to carry out unusual tasks. The behaviour of the participants may be distorted and not be like behaviour that would be carried out in the real world. Therefore, it should be difficult to generalise findings from experiments because they are not usually ecologically valid (true to real life).

A further difficulty with the experimental method is demand characteristics. Demand characteristics are all the cues which convey to the participant the purpose of the experiment. If a participant knows they are in an experiment they may seek cues about how they think they are expected to behave.

Another problem with the experimental method concerns ethics. For example, experiments often involve deceiving participants to some extent. However, it is possible to obtain a level of informed consent from participants. That is, the experimenter can provide information about what is going to happen without giving away the full aim of the study. This helps participants decide if they really want to take part.

It is recommended that participants in experiments are effectively debriefed and that the participants are clear that they can withdraw from the study at any time.

It is important to recognise that there are very many areas of human life which cannot be studied using the experimental method because it would be simply too unethical to do so.

 

Field experiments

A field experiment is an experiment that is conducted in ?the field ?. That is, in a real world situation. In field experiments the participants are not usually aware that that they are participating in an experiment.

The independent variable is still manipulated unlike in natural experiments. Field experiments are usually high in ecological validity and may avoid demand characteristics as the participants are unaware of the experiment.

 

However, in field experiments it is much harder to control confounding variables and they are usually time consuming and expensive to conduct.

In field experiments it is not usually possible to gain informed consent from the participants and it is difficult to debrief the participants.

 

Quasi or natural experiments

Quasi experiments are so called because they are not classed as true experiments.

A quasi experiment is where the independent variable is not manipulated by the researcher but occurs naturally. These experiments are often called natural experiments.

In a true experiment participants are allocated to the conditions of an experiment, usually through random assignment, however this is not always possible for practical or ethical reasons.

In a quasi experiment the researcher takes advantage of pre-existing conditions such as age, sex or an event that the researcher has no control over such as a participants? occupation.

A strength of some quasi experiments is that participants are often unaware that they are taking part in an investigation and they may not be as artificial as laboratory experiments.

However, it is argued that with quasi experiments it is harder to establish causal relationships because the independent variable is not being directly manipulated by the researcher.

It is worth noting that quasi experiments are very common in psychology because ethically and practically they are the only design that can be used.

 

Experimental Design

An important procedure to be aware of when researchers carry out experiments is experimental design.

An experimental design is a set of procedures used to control the influence of participant variables so that we can investigate the possible effects of the independent variable on the dependent variable. There are three basic experimental designs - independent measures design, repeated measures design and matched pairs design.

 

An independent measures design consists of using different participants for each condition of the experiment. If two groups in an experiment consist of different individuals then this is an independent measures design.

This type of design has an advantage resulting from the different participants used in each condition - there is no problem with order effects

The most serious disadvantage of independent measures designs is the potential for error resulting from individual differences between the groups of participants taking part in the different conditions. Also an independent groups design may represent an uneconomic use of those participants, since twice as many participants are needed to obtain the same amount of data as would be required in a two-condition repeated measures design.

 

A repeated measures design consists of testing the same individuals on two or more conditions.

The key advantage of the repeated measures design is that individual differences between participants are removed as a potential confounding variable. Also the repeated measures design requires fewer participants, since data for all conditions derive from the same group of participants.

The design also has its disadvantages. The range of potential uses is smaller than for the independent groups design. For example, it is not always possible to test the same participants twice.

There is also a potential disadvantage resulting from order effects, although these order effects can be minimised. Order effects occur when people behave differently because of the order in which the conditions are performed. For example, the participant?s performance may be enhanced because of a practice effect, or performance may be reduced because of a boredom or fatigue effect.

Order effects act as a confounding variable but can be reduced by using counterbalancing. If there are two conditions in an experiment the first participant can do the first condition first and the second condition second. The second participant can do the second condition first and the first condition second and so on. Therefore any order effects should be randomised.

 

A matched pairs design consists of using different participants for each condition of the experiment but participant variables are controlled by matching pairs of variables on a key variable.

In order to get the pairing precise enough, it is common to get one group of participants together and then look round for partners for everyone. Participants can be matched on variables which are considered to be relevant to the experiment in question. For example, pairs of participants might be matched for their scores from intelligence or personality tests.

Although this design combines the key benefits of both an independent and repeated measures design, achieving matched pairs of participants is a difficult and time consuming task which may be too costly to undertake. Successful use of a matched pairs design is heavily dependent on the use of reliable and valid procedures for pre-testing participants to obtain matched pairs.

 

 

Hypotheses

When carrying out experiments it is expected that the researcher will start with a hypothesis.

A hypothesis is a testable, predictive statement. The hypothesis will state what the researcher expects to find out. For example, participants who are tested at 10am will perform significantly better on a memory test than participants who are tested at 10pm.

It is important that the independent and dependent variables are clearly stated in the hypothesis.

When a hypothesis predicts the expected direction of the results it is referred to as a one-tailed hypothesis. For example the hypothesis above is stating that participants will perform better in the morning than the evening and is therefore a one-tailed hypothesis.

When a hypothesis does not predict the expected direction of the results it is referred to as a two-tailed hypothesis. For example a two tailed hypothesis might be that there will be a difference in performance on a memory test between participants who are tested at 10am and participants who are tested at 10pm

The hypothesis that states the expected results is called the alternate hypothesis because it is alternative to the null hypothesis. When conducting an experiment it is important that we have an alternate hypothesis and a null hypothesis. The null hypothesis is not the opposite of the alternate hypothesis it is a statement of no difference. A null hypothesis might be that there will be no significant difference on the performance on a memory test between participants who are tested at 10am and participants whom are tested at 10pm.

The reason we have a null hypothesis is that the statistical tests that we use are designed to test the null hypothesis.

 

More about extraneous variables and control

Extraneous variables are often classified as participant and situational variables.

When carrying out an experiment using an independent measures design it may be possible that participant differences are a confounding variable. For example if we find out that participants perform better on a test in a morning than an evening it may be that the participants who took the test in the morning are better at memory tests. An obvious way of controlling for participant variables is using a repeated measures design. Furthermore having a larger sample and randomly allocating participants to each condition may reduce participant variables. It may also be possible to use a matched pairs design where each participant could be matched with another in terms of their memory performance.

Situational variables are any feature of the experiment which could influence the participant?s behaviour. Experimenters attempt to control environmental factors such as noise by ensuring that these are consistent for both conditions. With a repeated measures design order effects can be controlled for by counterbalancing.

A way of reducing demand characteristics is to use a single blind technique whereby participants are not aware of the aims of the experiment. Furthermore, to reduce investigator or experimenter bias a double blind technique could be employed whereby both the participant and the researcher carrying out the experiment are unaware of the aim of the experiment.

 

Descriptive Statistics

Experiments produce quantitative data which can be analysed statistically. Statistics are a method of summarising and analysing data for the purpose of drawing conclusions about the data.

We can make a distinction between descriptive and inferential statistics.

Descriptive statistics simply offer us a way to describe a summary of our data.

Inferential statistics go a step further and allow us to make a conclusion related to our hypothesis. You may be pleased to know that we will not be doing inferential statistics until the second year.

Descriptive statistics give us a way to summarise and describe our data but do not allow us to make a conclusion related to our hypothesis.

When carrying out an experiment there are two main ways of summarising the data using descriptive statistics. The first way is to carry out of measure of central tendency (mean, median or mode) for each of the two conditions.

The mean is the arithmetic average that indicates the typical score in a data set and is calculated by adding all the scores together in each condition and then dividing by the number of scores. This is a useful statistic as it takes all of the scores into account but can be misleading if there are extreme values. For example if the scores on a memory test were 2, 4, 5, 6, 7, 42, the mean would be 10 which is not typical or representative of the data. The mean can not be used with nominal data. Nominal data are data in the form of separate categories such as grouping people according to their favourite type of cheese.

The median is calculated by finding the mid point in on ordered list. The median is calculated by placing all the values of one condition in order and finding the mid- point. This is a more useful measure than the mean when there are extreme values. For example, six scores on a test out of 100 are 70, 74, 75, 77, 78, 100. The mean is 79 but this is misleading in the sense that only one of the six participants has scored this high. The median score of 76 is a better description of the data. The disadvantage of the median is though that not all of the scores are taken into account. The median can also not be used when data are nominal.

The mode is the most common value in a set of values. The measure is used when we have nominal data such as number of people who prefer Mozzarella. When looking at a set of scores the mode is the score that applies to the greatest number of participants. For example, with scores of 30, 30, 30, 50, 96, 100 the mean is 61 which is misleading in the sense that no-one scored anywhere near this; the median is 40, which again does not approximate to anyone's score and the mode is 30, which at least lets us know that more people obtained this score than any other score. The mode is useful in certain instances where other measures of central tendency are rather meaningless. For instance, if you are a buyer for a shop whose target population consists of 50% of people who wear size 12 clothes and the remaining 50% are size 16 then it is no use you ordering size 14 clothes just because this is the average size.

Modes are used less often than other two measures of central tendency as they do not tell us anything about other scores in the distribution; they often are not very 'central' and they tend to fluctuate from one random sample of a population to another more often than either the median or the mean.

The second way of summarising and describing data is to calculate a measure of dispersion. This simply shows us the spread of a set of data.

A simple way of calculating the measure of dispersion is to calculate the range. The range is the difference between the smallest and largest value in a set of scores. Although it is a fairly crude measure of dispersion as any one high or low scale can distort the data. The range is usually used when we have used the median as the measure of central tendency.

A more sophisticated measure of dispersion is the standard deviation (SD) which tells us how much on average, scores differ from the mean. The standard deviation is usually used when we have used the mean as the measure of central tendency. The range only takes account of the highest and lowest scores but the SD takes every score into account. If the standard deviation of a group of scores is large, this means that the scores are widely distributed with many scores occurring a long way from the mean. If the standard deviation is small, most scores occur very close to the mean. Standard deviation can be defined as a statistical device for describing the variability of measurements.

Graphs

We can very easily create a graphical display of descriptive statistics. For example, bar charts can show at a glance the measures of central tendency of two conditions in an experiment. When drawing a bar chart ensure that it has a title and the axes are labelled. You can even colour the bars in.