FW: SPSS and Mode Averages

[Lemon, John S.]

I have used and supported SPSS since 1975 but due to rulings within the University the Computing Centre / IT Services were not allowed to give statistical advice so I have concentrated on data management and charts and graphs and let what little statistical knowledge I had lapse; so when  a user presented me with this problem I’m stuck. Please would someone be kind enough to explain the apparent, to him, discrepancies ? I assume it is to do with rounding during the calculations ?

 

I want to check how SPSS treats the mode average. I have found it can do this (via the frequencies) option but seems to consistently get it wrong – i.e. it always chooses the one after the mode (so if the mode – most frequent – answer was 0 it returns a 1). Compare the mode below for single authored articles with the frequencies data. I am worried the same error is creeping into the medians. For example, for co-authored articles I can’t see how the median (middle) entry could be 1 when 77% odd of the respondents answered 0.

 

Perhaps I’m misunderstanding the stats, perhaps there is something about the way SPSS is using the data, but I want to make sure there is not a generalised error inflating the stats (which could lead to serious egg on face later!)

 

Statistics
		How many single or co-authored journal articles have you ...	How many of these were published in journals outside of t...	How many co-authored journal publications have you publis...	How many articles did you publish prior to recieving an academic post with research in the contract?	What is the largest number of single and co-authored jour...
N	Valid	240	238	226	239	242
	Missing	14	16	28	15	12
Mean		4.66	2.57	1.38	3.85	3.20
Std. Error of Mean		.204	.127	.068	.187	.079
Median		4.00	2.00	1.00	3.00	3.00
Mode		3	1	1	1a	3
Std. Deviation		3.154	1.960	1.019	2.890	1.227
Skewness		2.353	2.178	4.235	1.293	2.241
Std. Error of Skewness		.157	.158	.162	.157	.156
Kurtosis		7.396	6.356	21.849	1.031	10.834
Std. Error of Kurtosis		.313	.314	.322	.314	.312
Minimum		1	1	1	1	1
Maximum		21	12	9	12	12
Sum		1118	612	313	921	775
a. Multiple modes exist. The smallest value is shown

 

 

How many single or co-authored journal articles have you ...
		Frequency	Percent	Valid Percent	Cumulative Percent
Valid	0	12	4.7	5.0	5.0
	1	29	11.4	12.1	17.1
	2	62	24.4	25.8	42.9
	3	51	20.1	21.3	64.2
	4	26	10.2	10.8	75.0
	5	18	7.1	7.5	82.5
	6	17	6.7	7.1	89.6
	7	2	.8	.8	90.4
	8	7	2.8	2.9	93.3
	9	2	.8	.8	94.2
	10	2	.8	.8	95.0
	11	3	1.2	1.3	96.3
	12	3	1.2	1.3	97.5
	14	2	.8	.8	98.3
	15	1	.4	.4	98.8
	16	1	.4	.4	99.2
	20	2	.8	.8	100.0
	Total	240	94.5	100.0
Missing	System	14	5.5
Total		254	100.0

 

 

How many co-authored journal publications have you publis...
		Frequency	Percent	Valid Percent	Cumulative Percent
Valid	0	176	69.3	77.9	77.9
	1	36	14.2	15.9	93.8
	2	6	2.4	2.7	96.5
	3	1	.4	.4	96.9
	4	3	1.2	1.3	98.2
	5	2	.8	.9	99.1
	6	1	.4	.4	99.6
	8	1	.4	.4	100.0
	Total	226	89.0	100.0
Missing	System	28	11.0
Total		254	100.0

 

 

 

Dr Andrew Dilley

Lecturer in History

http://www.abdn.ac.uk/history/staff/details.php?id=a.dilley

 

 

 

 

The University of Aberdeen is a charity registered in Scotland, No SC013683.

[Mark Miller]

John,

There is only one median -- it is unique.

There can be multiple modes, and SPSS tells you, for your data, 

that this is the case in a note at the bottom of the table.

... Mark Miller

On Tue, Mar 12, 2013 at 3:05 AM, Lemon, John <[hidden email]> wrote:

I have used and supported SPSS since 1975 but due to rulings within the University the Computing Centre / IT Services were not allowed to give statistical advice so I have concentrated on data management and charts and graphs and let what little statistical knowledge I had lapse; so when  a user presented me with this problem I’m stuck. Please would someone be kind enough to explain the apparent, to him, discrepancies ? I assume it is to do with rounding during the calculations ?

 

I want to check how SPSS treats the mode average. I have found it can do this (via the frequencies) option but seems to consistently get it wrong – i.e. it always chooses the one after the mode (so if the mode – most frequent – answer was 0 it returns a 1). Compare the mode below for single authored articles with the frequencies data. I am worried the same error is creeping into the medians. For example, for co-authored articles I can’t see how the median (middle) entry could be 1 when 77% odd of the respondents answered 0.

 

Perhaps I’m misunderstanding the stats, perhaps there is something about the way SPSS is using the data, but I want to make sure there is not a generalised error inflating the stats (which could lead to serious egg on face later!)

 

Statistics
		How many single or co-authored journal articles have you ...	How many of these were published in journals outside of t...	How many co-authored journal publications have you publis...	How many articles did you publish prior to recieving an academic post with research in the contract?	What is the largest number of single and co-authored jour...
N	Valid	240	238	226	239	242
	Missing	14	16	28	15	12
Mean		4.66	2.57	1.38	3.85	3.20
Std. Error of Mean		.204	.127	.068	.187	.079
Median		4.00	2.00	1.00	3.00	3.00
Mode		3	1	1	1a	3
Std. Deviation		3.154	1.960	1.019	2.890	1.227
Skewness		2.353	2.178	4.235	1.293	2.241
Std. Error of Skewness		.157	.158	.162	.157	.156
Kurtosis		7.396	6.356	21.849	1.031	10.834
Std. Error of Kurtosis		.313	.314	.322	.314	.312
Minimum		1	1	1	1	1
Maximum		21	12	9	12	12
Sum		1118	612	313	921	775
a. Multiple modes exist. The smallest value is shown

 

 

How many single or co-authored journal articles have you ...
		Frequency	Percent	Valid Percent	Cumulative Percent
Valid	0	12	4.7	5.0	5.0
	1	29	11.4	12.1	17.1
	2	62	24.4	25.8	42.9
	3	51	20.1	21.3	64.2
	4	26	10.2	10.8	75.0
	5	18	7.1	7.5	82.5
	6	17	6.7	7.1	89.6
	7	2	.8	.8	90.4
	8	7	2.8	2.9	93.3
	9	2	.8	.8	94.2
	10	2	.8	.8	95.0
	11	3	1.2	1.3	96.3
	12	3	1.2	1.3	97.5
	14	2	.8	.8	98.3
	15	1	.4	.4	98.8
	16	1	.4	.4	99.2
	20	2	.8	.8	100.0
	Total	240	94.5	100.0
Missing	System	14	5.5
Total		254	100.0

 

 

How many co-authored journal publications have you publis...
		Frequency	Percent	Valid Percent	Cumulative Percent
Valid	0	176	69.3	77.9	77.9
	1	36	14.2	15.9	93.8
	2	6	2.4	2.7	96.5
	3	1	.4	.4	96.9
	4	3	1.2	1.3	98.2
	5	2	.8	.9	99.1
	6	1	.4	.4	99.6
	8	1	.4	.4	100.0
	Total	226	89.0	100.0
Missing	System	28	11.0
Total		254	100.0

 

 

 

Dr Andrew Dilley

Lecturer in History

http://www.abdn.ac.uk/history/staff/details.php?id=a.dilley

 

 

 

 

The University of Aberdeen is a charity registered in Scotland, No SC013683.