Descriptive Statistics Lab

8/30/2001

Conti and Musty (1984) injected THC into the nucleus accumbens to see if that is the site at which THC has its effect. They expected that moderate doses of THC would increase an animal’s activity, while high doses would decrease it. They defined 5 experimental groups based on the amount of THC injected. These groups were

1 = Control (sham injection)
2 = 0.1 µg/kg THC
3 = 0.5 µg/kg THC
4 = 1.0 µg/kg THC
5 = 2.0 µg/kg THC

To quote from the text, "Conti and Musty took as their dependent variable the rat's activity for the 10 minutes after the injection as a proportion of the rat's activity in the 10 minutes before the injection. Since animals generally decrease their activity as they become accustomed to the apparatus, ratios were expected to be less than 1. However, it was anticipated that those rats with intermediate levels of THC would decrease their activity less (exhibit a higher post-injection/pre-injection ratio) than would those with either low or high levels. (Intermediate levels were expected to lead to the greatest activity, because very low doses should be insufficient to produce an effect and high doses should lead to decreased activity.)"

You must first enter the data to create a SPSS dataset

First start up SPSS

Then double click on the column headings, which will take you to a new view of the page. Then fill in the variable names.

Note: You only want two columns of data, not five. Column 1 will carry information on the Group membership, and Column 2 will carry information on the Activity score. Number the groups 1 - 5.

On the dialog box, chose Type and chose the number of digits and the number of decimals (0). That will make the data neater.

Now enter the data, either down the columns or across rows.

Next save the data as an SPSS dataset. I would name them something like Conti.sav, to indicate that they are an SPSS data set (as opposed to Conti.dat, which would imply an ASCII file) You are saving it simply to make life easier if your system crashes during the lab. We are likely to use these data again, so I would suggest saving them to zoo. You should probably set up a separate folder at zoo. I suggest a VERY consistent rule for capitalization, because Unix is case-sensitive.

Next you want to calculate the descriptive statistics.

Get the descriptive statistics for the dependent variable, ignoring group membership. But to make sure you have the correct data entered, your overall mean should come out to 46.00, with a standard deviation of 17.57.

I’m going to leave it up to you to figure out how to calculate the means for the five groups. Play around with the menus a bit and you should figure it out. There are actually 2 ways, but I prefer the way that will put all five means in the same table.

This procedure will only give means and variances, etc. It will not plot the data.

To get a histogram for the complete set of data, do the following.

1. Use the Graph menu to display a histogram of the dependent variable, ignoring Groups.
2. Then use the Data menu to split the file by Groups
3. Use the Histogram to give you the histograms for each group separately.
4. The list of output in the left margin will allow you to specify which output you want, or you can simply scroll the output screen.
5. Be sure to turn off the "Split File" selection before you do things that will require all 47 cases.
6. See if you can find a better way to plot the data so that you can see all the distributions at once. I did it using boxplots, but perhaps you can find another way.
7. For today, you don’t have to save these graphs. Just let them sit there. If you do want to save them, the dialogue box will let you save all at once, which is better than saving individual files.
8. Cut and paste those graphs to a Word document--I will go over that in class.
9.     a.  For a graph, you just "Copy" and "Paste."
10.     b.  For a table you need to "Copy Object". It does make a difference.

Draw a set of conclusions that Conti and Musty might have drawn, assuming that the major differences that you see are reliable.

Introini-Collison and McGaugh (1986) were interested in the effects of epinephrine on memory. They taught mice to run a simple Y maze with reinforcement on the left. Shortly thereafter they injected the mice with 0.0, 0.3, or 1.0 mg/kg of epinephrine (a chemical that circulates normally in the body, also known as adrenaline). They hypothesized that the medium dose would facilitate memory, and that the large dose would impair memory, both in comparison to the no-drug control. Sometime later the experimenters brought the mice back and trained them on a reversal, where the previously correct path is wrong, and the incorrect path right. Those who remember more from Day 1 should take longer to learn the reversal.

Here you are going to load data from an ASCII file. The file is named epineq.dat and is to be found on Gumby. (I will tell you how to get to that server.)  There are four columns of data there, so you will have to tell SPSS that—sort of. (The first column is an ID number. The second column contains the group number, defined by the Amount of the drug that was injected. The groups were  coded as 1, 2, 3 instead of 0, .3, 1. There was also another variable (the 3rd column) related to the Interval between learning and testing, and this is also coded as 1, 2, 3. We will read this variable in, since it is part of the data set, but we will then ignore it. The last column contains the trials to learn the reversal. ) I will go over how to read these data in the lab. The method changed between version 8 and version 9. The following should work for version 10.

From SPSS select Open from the File menu. Then find the file that you need.  (I know I’m leaving a lot for you to figure out, but I’ll be there to help if you can’t. I think you’ll learn more this way.) Click your way through the screens, but you probably will not need to change any settings. When you get to the appropriate screen, tell it that the first variable is SubjNum, the second is Amount, the third is Interval, and the fourth is DepVar (a higher score better memory of the original learning). 

When you have told it the four variable names, you are all set. Just press OK and it should load the data. Check over the spreadsheet that you will get to make sure that it makes sense (e.g. that it doesn’t have a random score down for some fifth variable, or an Amount score of 43).

Look at histograms and boxplots, and then calculate the relevant descriptive statistics. (For these, you can ignore the Interval variable completely.)

What should Introini-Collison and McGaugh conclude, assuming that the differences due to Amount are significant, but the differences due to Interval are not?

Write this up as if you were summarizing the study for a paper you were writing on variables relevant to memory. (In other words, saying "Group 1 had a mean of 3.67" is not remotely enough. You should briefly describe the study and the results in one paragraph—certainly no more than two. Hand it in at the next class. I do not expect that you will have run a statistical hypothesis test.)