Module 4-8 DQs

Posted: August 13th, 2013

Name:

Lecturer:

Course:
Date:

Module 4-8 DQs

Question One

a) Test retest reliability is used for assessing how consistent a test cross time is. This reliability type assumes that the construct or quality being measured does not go through any kind of change. Test retest reliability is best suited for things with stability such as intelligence. Primarily, reliability is higher when the time gap between tests is short. On the other hand, inter-rater reliability involves having two or more judges scoring the test. Consequently, these tests are subjected to comparison in order to establish how the rater estimates are consistent (Carmines, Edward, and Zeller, 38).

b) Content validity implies a non-statistical kind of validity that involves systematically examining the test material to establish whether it covers the sample behavior domain. The evidence on content validity includes the extent to which the test content matches a content domain. Tests have content validity built into them through careful selection terms such as compliance with certain test specifications. Predictive reliability refers to the extent where operationalization can correlate or predict other measures of similar construct to be measured in the future. The tests are administered, the performance is reviewed, and the scores are determined for correlation or prediction (Carmines, Edward, and Zeller, 74). Finally, construct validity is used to refer to whether a scale is able to correlate or measure with the theoretical scientific construct. Construct validity has a relation to the trait being considered, and is used as a means of assessing how that trait has been accomplished.

c) Internal validity involves conducting studies that establish casual relationships between variables. A study is regarded as having internal validity if it exhibits results that indicate changes in criterion or dependent variables occurred due to manipulation of the predictor or independent variables. Internal validity can also be considered as causal validity. In other words, it can show causality evidence. This is true for quantitative and qualitative designs. On the other hand, external validity involves the ability of the study to make the results general across experimental procedures, groups of people, or different situations (Krueger, Alan, and Schkade, 36). Primarily, internal validity is a way of stating that the study conducted is measuring what it is intended to examine. External validity shows how the findings are relevant in more than one situation, sample, or procedure. When a study is being designed, there is need to aim towards achieving good external and internal validity.

d) Indeed, a test can be reliable without being valid. Take for example a sample test undertaken to establish the notion that rate of intelligence is directly proportional to the size of a person’s head. Hence, this implies that a person’s intelligence becomes higher with a larger head size. Thus, the test is supposed to measure head circumference to determine the person’s intelligence. In reality, this would not be a reliable measure since most people would not agree because head circumference is not a valid way of measuring intelligence. However, reliability involves measuring consistency rather than truthfulness. In this regard, the above test is consistent when it comes to measuring head circumference and would not change with time. Hence, the score by an individual would still be equal or slightly different to that of another, thus the test would become reliable. However, the test is not a valid intelligence measure since it lacks validity.

Question Two

The components that differentiate experiments and non-experiments are based on several factors. These include random sampling. In this case, an experiment method involves random sampling where each set has an equal and independent opportunity of being selected. A non-experiment method is contrary to this. A non-experiment experiment method may not induce a random system of assigning subjects into the treatment or control group. The members of the set in this case do not have an equal chance of being included in the treatment or control group. Furthermore, an experiment method practices experiment manipulation and control. This involves directly manipulating variables for testing cause and effect relationships (Salkind, Neil , and Rasmussen, 67). On the other hand, non-experiment methods do not take full control of all extraneous conditions and variables that have an influence on the outcome of the experiment. Experiments do a good job of assessing causality through drawing conclusions regarding the variables’ causal relationships. Additionally, these experiments give a credible view regarding what occurred in the course of the experiment.

Question Three

a) For the first experiment, the design includes randomly assigned experiments where only one variable can be tested and manipulated. The tested subjects have to be assigned to experimental groups or either control. On the other hand, the scientific control group should be used to ascertain that the design of the experiment can generate the necessary results.

b) If the experiment involves three independent variables with three levels, then it should be conducted in the form of a single factor design with three levels. For example, if three dosages of medication are available, then 1/3 of the population would receive a single dose. On the other hand, if the same design is within three independent variables with two levels, then ½ of the population would receive the medication.

Question Four

Between subjects designs are fundamental in many situations by giving researchers an opportunity of conducting the experiment with few extraneous factors or little contamination. Additionally, this design is more independent since every participant is subjected to one treatment each (Krueger, Alan, and Schkade, 54). Hence, this reduces the chances of boredom arising among the participants after several test series. Additionally, it allows the participants to become more accomplished through practicing and experience, hence skewing the outcome of the experiment.

Question Five

a) A within subjects design involves conducting a type of design experiment that exposes all participants to every condition or treatment. The term treatment is supposed to imply the different levels associated with the independent variables. The other scenario is a between subjects design and this involves testing all the participants with regard to a particular condition. The first group is tested in accordance with the condition A, the second group with condition B respectively. The assumption here maintains that conditions A and B are of the same factor but have different levels.

b) Large N designs draw their conclusions by comparing between substantial numbers and conditions of subjects whereas small N designs focus on how an individual or small group behaves. The large N design groups participants and treats them wholly rather than individually. On the other hand, small N designs present individual data (Salkind, Neil, and Rasmussen, 92).

c) I would use the within subjects design in circumstances where effect sizes are small and extraneous variables involving subject characteristics will interfere with the relationships of cause and effect. On the other hand, I would opt to use the large N design when I intend to compare how the groups performed on certain subjects, and when I am about to test competing hypothesis as well as examining to what extent the findings are general.

Question Six

Fatigue and boredom can be reduced within subject designs by administering only two or one treatments in every session. Additionally, the sessions would be spread over on a weekly basis. However, if this is the approach employed, it is important to understand that it will mostly likely exhibit subject attrition. Furthermore, if the subject does not show up for a particular session, the subject data will have to be discarded since considerable time will have been wasted.

Question Seven

Given the following students’ test scores (95, 92, 90, 90, 83, 83, 83, 74, 60, and 50), the mean would be calculated as follows. Add up the entire test scores to and divide the sum with the number of participants. The sum adds up to 717 and dividing it by 10 gives a mean of 71.7 points. The median value 83, the mode is 83 since it is the most common.

Range is equal to 95 – 74= 21, the standard deviation is 65.85, and the variance adds up to 76.75. In this regard, the information highlighted above would be useful in the field of statistics and data analysis. It can also be useful in analyzing and interpreting results, explaining data variations, and predicting future data.

Question Eight

a) Null hypothesis is that kind of hypothesis that gives particular information regarding a particular population parameter. The purpose of this testing involves testing how viable the null hypothesis is when the experimental data is taken into consideration. The null hypothesis is a reverse of what the person undertaking the experiment believes. It is conducted with the purpose of putting forward the established data and consequently attempting to contradict it (Carmines, Edward, and Zeller, 104).

b) In this regard, a null hypothesis includes all default statistical hypothesis or original hypothesis. On the other hand, alternative hypothesis is all other hypothesis other than null. Hence, if the null hypothesis is not taken into consideration, then alternative hypothesis is taken. This makes them both necessary in statistics.

Question Nine

a) The results of the test with 38 degrees of freedom would be shown in a table. This table would exhibit the performed statistical analyses and the hypothesis tested. Hence for the t test conducted, I would report the t statistic and p value without considering whether the p-value has any statistical significance. Hence, one sample t(38) = p value .20

b) The proper placement of figure within the text would involve replacing “df”, “p-value” and “t-value” with their measured values. Regarding the report digit numbers, our main concern rests with whether p has a bigger value than 20.

c) The proper reference for the placing a table within the text will involve assigning each cell with the respective freedom subject and the number of respondents that favored that particular subject.

d) The proper reference for the textbook would be done in accordance with the specified guidelines such as APA, MLA, or Chicago rules.

Question Ten

Increasing internal and external validity in an experiment can be achieved through randomization, using statistical and research analysis that is appropriate to the data types collected. Using experiments with single subjects has a higher internal and external validity since the subjects are their own controls.