Dream Guide Team
10-03-2008, 10:17 AM
Written by ninja9578
Dreamviews Research Guidelines
The Dreamviews community has a lot of good ideas for ways to help themselves and others become lucid and control their dreams. Unfortunately most of these ideas go untested; the community accepts them at face value without going through the proper steps to see whether or not they are actually as useful as they seem.
The problem with this area of psychology is that it's intangible, there is no mathematical proof for anything. Things like placebo, unnecessary steps or ideas, and just flat out jests are prevalent on the forum. These variables can cause mass confusion on which concepts are concrete truth, and which are just useless fluff.
Proper care must be taken when testing the objective validity of any claims or commonly accepted beliefs. The following is a guideline on what is expected for an idea to be considered either proven or disproven.
The Scientific Method.
All science must abide the rules of the scientific method. The scientific method is comprised of these steps:
Hypothesis - Coming up with an idea and explaining why you think it will work
. Gathering Data - Gathering all necessary data, including all variables
. Analyzing Data - Fitting data to accepted methods of analysis
. Accept or Reject - your hypothesis is either true or false, no middle ground. If the hypothesis is rejected, then it is modified and the process is repeated.
Hypothesis.
Your hypothesis should be something that is either true or false. There should be no middle ground possible. "I think WILD works" is not a scientific hypothesis. It must be modified to something that is either true or false. "I think WILD will produce LDs in experienced dreamers 50%" would be a more scientific way of forming your hypothesis.
You must also lay out in detail what you mean by WILD. You have to control all variables such as what time the participants get up to do it, how long then can try before considering it a failure...
Experiment.
When designing and proposing your experiment, you want to find a way to weed out as many variables as possible. This means that you may want to allow everyone the same amount of sleep, and maybe taking the same lucid aids.
Double Blind Method
If possible, the double blind method of research is preferred. This will mostly be applicable for lucid aids, but you want to have two groups for your experiments, a control group, who gets a placebo, and an experiment group that gets the real thing. This may prove difficult since you are not administering the test yourself.
For a lucid supplement experiment, you may want to dreamer to place two pills in a bottle, shake it up and then grab and take one pill without looking. The next morning they can look to see which pill was still in the bottle and record which one they had taken: the placebo or the real thing.
Gathering Data.
Selecting a Sample
For our purposes here at DV, it is usually acceptable to ask for volunteers to do it. You have to gather information about each participant though. This includes all relevant data: baseline LD rate, experience level, hours of sleep. These are just examples, specific data will be required for each experiment. For instance, for a WILD study, it may be necessary to get their WILDing experience level, what time they get up to WILD, what position they use to WILD...
The larger the sample size the better, statisticians recommend 40 subjects to properly create a normal curve.
.The Data
It is recommended to collect data every day. By collecting data every day it is easier to see trends quicker and it may help you spots bugs in your experiment plan. This data can either be PM'd to you by participants or posted within the thread. To prevent a confidence boost or lull coming from other members seeing results before experimenting, PMs are recommended. However, if you contact a moderator the posts can be hidden from public view until you are done with the study.
Remember to take in not just lucid dream data, but other variables as well. This includes hours of sleep, any drugs that were taken, what they had for dinner (yeah, seriously), number of times they woke up during the night...
Analyzing Data.
Normal Distribution
The only way for a claim to be taken seriously is if many people experience that same result. A common technique used in statistics is using a normal curve to see probability density.
Here is how this kind of thing can work for us. Say you were doing a study on an induction technique and got this result.
Number of Lucid Dreams per person (in %)
00 ❚❚❚❚❚❚❚❚❚❚
10 ❚❚❚
20 ❚
30 ❚❚
40 ❚❚❚❚❚❚❚
50 ❚❚❚❚
60 ❚❚
70 ❚
80
90
Here we do not see a standard normal curve, we have a bimodal curve. We do not want this, we want a normal curve, so you have to explain why we have a bimodal curve and find a way to make it normal by looking at our other controls.
You can not claim that your technique produces lucid dreams, while it did for some members, it did nothing for others. Remember, there are a lot of factors and we have to look at all of them. We must separate the participants by things like experience level, baseline LD rate, hours of sleep...
Baseline LD rate is usually going to be the largest factor. You can change the curve to change in rate of lucid dreams.
Now we may get something like this.
Increase in Number of Lucid Dreams (in %)
-40
-30
-20❚
-10❚❚❚
00 ❚❚❚❚❚❚❚❚❚❚❚
10 ❚❚❚❚❚❚❚
20 ❚❚❚❚❚
30 ❚❚
40 ❚
Now we have a normal curve and can use our results. In the previous graph, it is likely that most of the members who had 0 lucid dreams with the method had been having 0 lucid dreams anyway, so that gave us skewed results. In this new graph compensating for baseline rate, we see that the mean is on the plus side. Therefore this technique gave an overall increase in the number of lucid dreams.
You must report all results, however, not just the second graph. The fictional technique here seemed to do very little for newbies and would therefore not be a good technique to start with. You should probably also separate the original graph into two groups, newbies and experienced dreamers. See correlation there.
.Standard Deviation
You also want to take a look at your standard deviation. This corresponds to how flat the curve is, a steeper curve is a better result, it means that everyone had fairly uniform results. If you have a flat curve then it means that there was a huge variation in results. This means that even if you had a positive result, it is not reliable.
Think of it like a game of darts. If you place the darts all over the place then the average of all of your throws may be very close to the bulls eye, but you never actually hit it. This is why averaging things without taking the standard deviation is not proper science.
The standard deviation is calculated by taking the mean value and calculating how far away every data point is away from that. Those values are each squared and then added up. Then taking the square root of that sum.
If the random variable X takes on N values (which are real numbers) with equal probability, then its standard deviation σ can be calculated as follows:
Find the mean, http://upload.wikimedia.org/math/8/c/a/8ca96727b4c0d19c76963a3d1b9488bd.png, of the values.
For each value xi calculate its deviation (http://upload.wikimedia.org/math/1/3/b/13bb700f5a5d221e89199af8874c8560.png) from the mean.
Calculate the squares of these deviations.
Find the mean of the squared deviations. This quantity is the variance σ2.
Take the square root of the variance.
This calculation is described by the following formula:
http://upload.wikimedia.org/math/6/1/6/616b3bfb2e10653483fa05b3ecab46ef.png
.Flowing Graphs
For some techniques it is important to show how the technique works over time. Some technique may take a while to take effect, some may take effect right away, but then stop working.
Each of the graphs mentioned above should have subgraphs to show change over time. This is particularly important when studying a technique that requires a change in sleeping habits. For WILDs, people not used to waking up in the middle of the night may take a while to show results because their bodies have to adjust to waking up and going back to sleep. It may also be likely that timing the technique will need fine tuning for each subject.
Some people may go into an experiment very excited and expecting to lucid dream, and therefore do. This may have little to do with the technique, but more the effect of thinking about LDing because of the study. Watching for effectiveness tapering off is important to claiming that the technique works.
The best way to show this is to plot a line graph with the mean and standard deviations, but remember to show hard data too. You can use the tags here at DV to hide the hard data and just show the line graphs for casual reviewers.
.
Confidence Interval
Confidence levels tell the readers of the study how sure you are that your study results represent the entire population. There are over 6 billion people on earth, there is no way to test your experiment for all of them. Confidence level is how sure you are that your results represent the average population.
Your confidence interval is whatever you wish to report it to be. Say that your supplement or technique in question worked on average 40% of the time. You can report that it will work 40% of the time, because some people had it work more, other had it work less. You want to give a range. Usually you will want to have a 95% confidence level that the range that you give accurately depicts the entire population. The actually calculations for finding your confidence interval is complicated and involve lookup charts so you can use this calculator to find it.
http://www.gifted.uconn.edu/siegle/research/Samples/samplecalculator.htm
Say for the experiment where you get 40% average and the calculator gives you a confidence interval of 21%. You must report that it will work between 19% and 61% of the time with a 95% confidence level.
Accepting or Rejecting.
Your data will either support your hypothesis or not. In order for you hypothesis to be correct the data has to fit your hypothesis within a statistically significant amount. There can be outliers to your data, however they must within three standard deviations or less away from the mean.
If more than 10% of your data does not fit your your hypothesis then it needs to be adjusted. If your hypothesis can be adjusted to fit your current data that's okay, but if you need to change your technique at all, then new data must be collected.
.What to Report
All induction technique should include at minimum these results:
Change in LD Rate
Change in LD Rate (just newbies)
Change in LD Rate (just experienced)
Change in LD Rate Flowing Graph
Change in LD Rate Flowing Graph (just newbies)
Change in LD Rate Flowing Graph (just experienced)
Each should include a graph, a mean, and a standard deviation
I recommend using the vertical bar graph using the black boxes like I did. This allows members reading your research to know exactly how many people participated in the study.
Report failed experiments too. This prevents others from making the same mistake. It also allows others to look at your work and, if possible modify the hypothesis to them to work. Science is rarely made up of successful experiments, but instead failed experiments. Thomas Edison was once quoted saying "I've never failed, I only invented 1000 ways not to make a lightbulb." His point was that a failed experiment is successful science.
Proposal Method.
When making a proposal for an experiment in the research forum you want to include your hypothesis, your experiment, and any other data needed. This may include links to research by other people, references to medical dictionaries, et cetera.
The proposed thread has to be approved by a moderator before it is posted. In some cases the proposed experiment will be accepted right away, other times the moderator may wish to ask for adjustments. There may be various reasons for this such as: you overlooked something important or you didn't follow the scientific method.
In the case of information that you do not want your subjects to know (such as for a double blind study) then you will receive a PM regarding whether or not your proposal was accepted. Then you may post another thread that the public will see.
Dreamviews Research Guidelines
The Dreamviews community has a lot of good ideas for ways to help themselves and others become lucid and control their dreams. Unfortunately most of these ideas go untested; the community accepts them at face value without going through the proper steps to see whether or not they are actually as useful as they seem.
The problem with this area of psychology is that it's intangible, there is no mathematical proof for anything. Things like placebo, unnecessary steps or ideas, and just flat out jests are prevalent on the forum. These variables can cause mass confusion on which concepts are concrete truth, and which are just useless fluff.
Proper care must be taken when testing the objective validity of any claims or commonly accepted beliefs. The following is a guideline on what is expected for an idea to be considered either proven or disproven.
The Scientific Method.
All science must abide the rules of the scientific method. The scientific method is comprised of these steps:
Hypothesis - Coming up with an idea and explaining why you think it will work
. Gathering Data - Gathering all necessary data, including all variables
. Analyzing Data - Fitting data to accepted methods of analysis
. Accept or Reject - your hypothesis is either true or false, no middle ground. If the hypothesis is rejected, then it is modified and the process is repeated.
Hypothesis.
Your hypothesis should be something that is either true or false. There should be no middle ground possible. "I think WILD works" is not a scientific hypothesis. It must be modified to something that is either true or false. "I think WILD will produce LDs in experienced dreamers 50%" would be a more scientific way of forming your hypothesis.
You must also lay out in detail what you mean by WILD. You have to control all variables such as what time the participants get up to do it, how long then can try before considering it a failure...
Experiment.
When designing and proposing your experiment, you want to find a way to weed out as many variables as possible. This means that you may want to allow everyone the same amount of sleep, and maybe taking the same lucid aids.
Double Blind Method
If possible, the double blind method of research is preferred. This will mostly be applicable for lucid aids, but you want to have two groups for your experiments, a control group, who gets a placebo, and an experiment group that gets the real thing. This may prove difficult since you are not administering the test yourself.
For a lucid supplement experiment, you may want to dreamer to place two pills in a bottle, shake it up and then grab and take one pill without looking. The next morning they can look to see which pill was still in the bottle and record which one they had taken: the placebo or the real thing.
Gathering Data.
Selecting a Sample
For our purposes here at DV, it is usually acceptable to ask for volunteers to do it. You have to gather information about each participant though. This includes all relevant data: baseline LD rate, experience level, hours of sleep. These are just examples, specific data will be required for each experiment. For instance, for a WILD study, it may be necessary to get their WILDing experience level, what time they get up to WILD, what position they use to WILD...
The larger the sample size the better, statisticians recommend 40 subjects to properly create a normal curve.
.The Data
It is recommended to collect data every day. By collecting data every day it is easier to see trends quicker and it may help you spots bugs in your experiment plan. This data can either be PM'd to you by participants or posted within the thread. To prevent a confidence boost or lull coming from other members seeing results before experimenting, PMs are recommended. However, if you contact a moderator the posts can be hidden from public view until you are done with the study.
Remember to take in not just lucid dream data, but other variables as well. This includes hours of sleep, any drugs that were taken, what they had for dinner (yeah, seriously), number of times they woke up during the night...
Analyzing Data.
Normal Distribution
The only way for a claim to be taken seriously is if many people experience that same result. A common technique used in statistics is using a normal curve to see probability density.
Here is how this kind of thing can work for us. Say you were doing a study on an induction technique and got this result.
Number of Lucid Dreams per person (in %)
00 ❚❚❚❚❚❚❚❚❚❚
10 ❚❚❚
20 ❚
30 ❚❚
40 ❚❚❚❚❚❚❚
50 ❚❚❚❚
60 ❚❚
70 ❚
80
90
Here we do not see a standard normal curve, we have a bimodal curve. We do not want this, we want a normal curve, so you have to explain why we have a bimodal curve and find a way to make it normal by looking at our other controls.
You can not claim that your technique produces lucid dreams, while it did for some members, it did nothing for others. Remember, there are a lot of factors and we have to look at all of them. We must separate the participants by things like experience level, baseline LD rate, hours of sleep...
Baseline LD rate is usually going to be the largest factor. You can change the curve to change in rate of lucid dreams.
Now we may get something like this.
Increase in Number of Lucid Dreams (in %)
-40
-30
-20❚
-10❚❚❚
00 ❚❚❚❚❚❚❚❚❚❚❚
10 ❚❚❚❚❚❚❚
20 ❚❚❚❚❚
30 ❚❚
40 ❚
Now we have a normal curve and can use our results. In the previous graph, it is likely that most of the members who had 0 lucid dreams with the method had been having 0 lucid dreams anyway, so that gave us skewed results. In this new graph compensating for baseline rate, we see that the mean is on the plus side. Therefore this technique gave an overall increase in the number of lucid dreams.
You must report all results, however, not just the second graph. The fictional technique here seemed to do very little for newbies and would therefore not be a good technique to start with. You should probably also separate the original graph into two groups, newbies and experienced dreamers. See correlation there.
.Standard Deviation
You also want to take a look at your standard deviation. This corresponds to how flat the curve is, a steeper curve is a better result, it means that everyone had fairly uniform results. If you have a flat curve then it means that there was a huge variation in results. This means that even if you had a positive result, it is not reliable.
Think of it like a game of darts. If you place the darts all over the place then the average of all of your throws may be very close to the bulls eye, but you never actually hit it. This is why averaging things without taking the standard deviation is not proper science.
The standard deviation is calculated by taking the mean value and calculating how far away every data point is away from that. Those values are each squared and then added up. Then taking the square root of that sum.
If the random variable X takes on N values (which are real numbers) with equal probability, then its standard deviation σ can be calculated as follows:
Find the mean, http://upload.wikimedia.org/math/8/c/a/8ca96727b4c0d19c76963a3d1b9488bd.png, of the values.
For each value xi calculate its deviation (http://upload.wikimedia.org/math/1/3/b/13bb700f5a5d221e89199af8874c8560.png) from the mean.
Calculate the squares of these deviations.
Find the mean of the squared deviations. This quantity is the variance σ2.
Take the square root of the variance.
This calculation is described by the following formula:
http://upload.wikimedia.org/math/6/1/6/616b3bfb2e10653483fa05b3ecab46ef.png
.Flowing Graphs
For some techniques it is important to show how the technique works over time. Some technique may take a while to take effect, some may take effect right away, but then stop working.
Each of the graphs mentioned above should have subgraphs to show change over time. This is particularly important when studying a technique that requires a change in sleeping habits. For WILDs, people not used to waking up in the middle of the night may take a while to show results because their bodies have to adjust to waking up and going back to sleep. It may also be likely that timing the technique will need fine tuning for each subject.
Some people may go into an experiment very excited and expecting to lucid dream, and therefore do. This may have little to do with the technique, but more the effect of thinking about LDing because of the study. Watching for effectiveness tapering off is important to claiming that the technique works.
The best way to show this is to plot a line graph with the mean and standard deviations, but remember to show hard data too. You can use the tags here at DV to hide the hard data and just show the line graphs for casual reviewers.
.
Confidence Interval
Confidence levels tell the readers of the study how sure you are that your study results represent the entire population. There are over 6 billion people on earth, there is no way to test your experiment for all of them. Confidence level is how sure you are that your results represent the average population.
Your confidence interval is whatever you wish to report it to be. Say that your supplement or technique in question worked on average 40% of the time. You can report that it will work 40% of the time, because some people had it work more, other had it work less. You want to give a range. Usually you will want to have a 95% confidence level that the range that you give accurately depicts the entire population. The actually calculations for finding your confidence interval is complicated and involve lookup charts so you can use this calculator to find it.
http://www.gifted.uconn.edu/siegle/research/Samples/samplecalculator.htm
Say for the experiment where you get 40% average and the calculator gives you a confidence interval of 21%. You must report that it will work between 19% and 61% of the time with a 95% confidence level.
Accepting or Rejecting.
Your data will either support your hypothesis or not. In order for you hypothesis to be correct the data has to fit your hypothesis within a statistically significant amount. There can be outliers to your data, however they must within three standard deviations or less away from the mean.
If more than 10% of your data does not fit your your hypothesis then it needs to be adjusted. If your hypothesis can be adjusted to fit your current data that's okay, but if you need to change your technique at all, then new data must be collected.
.What to Report
All induction technique should include at minimum these results:
Change in LD Rate
Change in LD Rate (just newbies)
Change in LD Rate (just experienced)
Change in LD Rate Flowing Graph
Change in LD Rate Flowing Graph (just newbies)
Change in LD Rate Flowing Graph (just experienced)
Each should include a graph, a mean, and a standard deviation
I recommend using the vertical bar graph using the black boxes like I did. This allows members reading your research to know exactly how many people participated in the study.
Report failed experiments too. This prevents others from making the same mistake. It also allows others to look at your work and, if possible modify the hypothesis to them to work. Science is rarely made up of successful experiments, but instead failed experiments. Thomas Edison was once quoted saying "I've never failed, I only invented 1000 ways not to make a lightbulb." His point was that a failed experiment is successful science.
Proposal Method.
When making a proposal for an experiment in the research forum you want to include your hypothesis, your experiment, and any other data needed. This may include links to research by other people, references to medical dictionaries, et cetera.
The proposed thread has to be approved by a moderator before it is posted. In some cases the proposed experiment will be accepted right away, other times the moderator may wish to ask for adjustments. There may be various reasons for this such as: you overlooked something important or you didn't follow the scientific method.
In the case of information that you do not want your subjects to know (such as for a double blind study) then you will receive a PM regarding whether or not your proposal was accepted. Then you may post another thread that the public will see.